Solving Interoperability in Translational Health. Perspectives of Students from the International Partnership in Health Informatics Education (IPHIE) 2016 Master Class.

BACKGROUND: In the summer of 2016 an international group of biomedical and health informatics faculty and graduate students gathered for the 16th meeting of the International Partnership in Health Informatics Education (IPHIE) masterclass at the University of Utah campus in Salt Lake City, Utah. This international biomedical and health informatics workshop was created to share knowledge and explore issues in biomedical health informatics (BHI). OBJECTIVE: The goal of this paper is to summarize the discussions of biomedical and health informatics graduate students who were asked to define interoperability, and make critical observations to gather insight on how to improve biomedical education. METHODS: Students were assigned to one of four groups and asked to define interoperability and explore potential solutions to current problems of interoperability in health care. RESULTS: We summarize here the student reports on the importance and possible solutions to the "interoperability problem" in biomedical informatics. Reports are provided from each of the four groups of highly qualified graduate students from leading BHI programs in the US, Europe and Asia. CONCLUSION: International workshops such as IPHIE provide a unique opportunity for graduate student learning and knowledge sharing. BHI faculty are encouraged to incorporate into their curriculum opportunities to exercise and strengthen student critical thinking to prepare our students for solving health informatics problems in the future.

Assessing Metadata Quality of a Federally Sponsored Health Data Repository.

The U.S. Federal Government developed HealthData.gov to disseminate healthcare datasets to the public. Metadata is provided for each datasets and is the sole source of information to find and retrieve data. This study employed automated quality assessments of the HealthData.gov metadata published from 2012 to 2014 to measure completeness, accuracy, and consistency of applying standards. The results demonstrated that metadata published in earlier years had lower completeness, accuracy, and consistency. Also, metadata that underwent modifications following their original creation were of higher quality. HealthData.gov did not uniformly apply Dublin Core Metadata Initiative to the metadata, which is a widely accepted metadata standard. These findings suggested that the HealthData.gov metadata suffered from quality issues, particularly related to information that wasn't frequently updated. The results supported the need for policies to standardize metadata and contributed to the development of automated measures of metadata quality.

Indexing Publicly Available Health Data with Medical Subject Headings (MeSH): An Evaluation of Term Coverage.

As part of the Open Government Initiative, the United States federal government published datasets to increase collaboration, transparency, consumer participation, and research, and are available online at HealthData.gov. Currently, HealthData.gov does not adequately support the accessibility goal of the Open Government Initiative due to issues of retrieving relevant data because of inadequately cataloguing and lack of indexing with a standardized terminology. Given the commonalities between the HealthData.gov and MEDLINE metadata, Medical Subject Headings (MeSH) may offer an indexing solution, but there needs to be a formal evaluation of the efficacy of MeSH for covering the dataset concepts. The purpose of this study was to determine if MeSH adequately covers the HealthData.gov concepts. The noun and noun phrases from the HealthData.gov metadata were extracted and mapped to MeSH using MetaMap. The frequency of no exact, partical and no matches with MeSH terms were determined. The results of this study revealed that about 70% of the HealthData.gov concepts partially or exactly matched MeSH terms. Therefore, MeSH may be a favorable terminology for indexing the HealthData.gov datasets.

Partnership to promote interprofessional education and practice for population and public health informatics: A case study.

Team-based healthcare delivery models, which emphasize care coordination, patient engagement, and utilization of health information technology, are emerging. To achieve these models, expertise in interprofessional education, collaborative practice across professions, and informatics is essential. This case study from informatics programs in the Academic Health Center (AHC) at the University of Minnesota and the Office of Health Information Technology (OHIT) at the Minnesota Department of Health presents an academic-practice partnership, which focuses on both interprofessionalism and informatics. Outcomes include the Minnesota Framework for Interprofessional Biomedical Health Informatics, comprising collaborative curriculum development, teaching and research, practicums to promote competencies, service to advance biomedical health informatics, and collaborative environments to facilitate a learning health system. Details on these Framework categories are presented. Partnership success is due to interprofessional connections created with emphasis on informatics and to committed leadership across partners. A limitation of this collaboration is the need for formal agreements outlining resources and roles, which are vital for sustainability. This partnership addresses a recommendation on the future of interprofessionalism: that both education and practice sectors be attuned to each other's expectations and evolving trends. Success strategies and lessons learned from collaborations, such as that of the AHC-OHIT that promote both interprofessionalism and informatics, need to be shared.

Traumatic brain injury rehabilitation: analysis of common data elements.

Physical medicine rehabilitation interventions for post-acute traumatic brain injury (TBI) are heterogeneous and subject to differences based on multi-disciplinary treatment plans [1]. There is no universal knowledge representation (KR) model for TBI rehabilitation which impedes data collection, aggregation, computation, and sharing. This paper describes results of an analysis of the National Institute for Neurological Disorders and Stroke (NINDS) TBI "Common Data Elements" (CDE) clinical data standardization set. We conducted this to understand current TBI rehabilitation KR and as a foundational step toward the creation of a domain ontology. A content coverage study was performed on the "Treatment/Intervention" sub-set of CDEs. Results show that coverage of the CDEs is broad but lacks depth to represent the context of data collection in the TBI rehabilitation process. Next steps will be development of a KR model and identification and validation of domain concepts for a foundational ontology.

Patterns of tooth loss in older adults with and without dementia: a retrospective study based on a Minnesota cohort.

OBJECTIVES: To study tooth loss patterns in older adults with dementia. DESIGN: Retrospective longitudinal study. SETTING: A community-based geriatric dental clinic in Minnesota. PARTICIPANTS: Four hundred ninety-one older adults who presented to the study clinic as new patients during the study period, remained dentate after finishing the initial treatment plan, and returned for care at least once thereafter were retrospectively selected. One hundred nineteen elderly people with International Classification of Diseases, Ninth Revision, codes 290.x, 294.1, or 331.2 or a plain-text diagnosis of dementia, Alzheimer's disease, or chronic brain syndrome in the medical history were considered having dementia. INTERVENTION: All existing dental conditions were treated before enrollment. Dental treatment was continually provided for all participants during follow-up. MEASUREMENTS: Tooth loss patterns, including time to first tooth loss, number of tooth loss events, and number of teeth lost per patient-year were estimated and compared for participants with and without dementia using Cox, Poisson, and negative-binomial regressions. RESULTS: Participants with dementia arrived with an average of 18 and those without dementia with an average of 20 teeth; 27% of remaining teeth in the group with dementia were decayed or retained roots, higher than in the group without dementia (P<.001). Patterns of tooth loss did not significantly differ between the two groups; 11% of participants in both groups had lost teeth by 12 months of follow-up. By 48 months, 31% of participants without dementia and 37% of participants with dementia had lost at least one tooth (P=.50). On average, 15% of participants in both groups lost at least one tooth each year. Mean numbers of teeth lost in 5 years were 1.21 for participants with dementia and 1.01 for participants without dementia (P=.89). CONCLUSION: Based on data available in a community-based geriatric dental clinic, dementia was not associated with tooth loss. Although their oral health was poor at arrival, participants with dementia maintained their dentition as well as participants without dementia when dental treatment was provided.

Predicting tooth loss for older adults with special needs.

OBJECTIVES: Older Adults with Special Needs (OASN) have more oral health needs compared with healthy, independent elders. Currently, little is known about tooth loss, a key indicator of oral function loss, among OASN. Risk assessment is primarily based on clinical experience rather than scientific evidence, raising concerns for quality of care. The study's objective was to develop an evidence-based model to quantitatively predict tooth loss for OASN. METHODS: Four hundred ninety-one dentate older adults, including 235 from long-term care facilities, were retrospectively recruited. Subjects were treated and brought to a state of oral health before enrollment. Medical and dental assessments were abstracted from dental records and used to predict risk of tooth loss. Tooth loss events were recorded for subjects during follow-up. Multivariate negative-binomial regression was used, starting with 27 risk factors and removing variables using Akaike's Information Criterion. Pearson's correlation was then conducted to evaluate the overall fit of the final fitted model. RESULTS: The final fitted model included eight predictors. Among them, age, number of decayed/broken teeth at arrival, anticholinergic burden of medications and physical mobility were associated with risk of tooth loss in OASN (P ≤ 0.05). Internal validation indicated satisfactory fit of the final fitted model. CONCLUSION: An evidence-based model with eight predictors was developed to quantitatively predict risk of tooth loss for OASN at the individual level.

Evaluation and optimization of fMRI single-subject processing pipelines with NPAIRS and second-level CVA.

In functional magnetic resonance imaging (fMRI) analysis, although the univariate general linear model (GLM) is currently the dominant approach to brain activation detection, there is growing interest in multivariate approaches such as principal component analysis, canonical variate analysis (CVA), independent component analysis and cluster analysis, which have the potential to reveal neural networks and functional connectivity in the brain. To understand the effect of processing options on performance of multivariate model-based fMRI processing pipelines with real fMRI data, we investigated the impact of commonly used fMRI preprocessing steps and optimized the associated multivariate CVA-based, single-subject processing pipelines with the NPAIRS (nonparametric prediction, activation, influence and reproducibility resampling) performance metrics [prediction accuracy and statistical parametric image (SPI) reproducibility] on the Fiswidgets platform. We also compared the single-subject SPIs of univariate GLM with multivariate CVA-based processing pipelines from SPM, FSL.FEAT, NPAIRS.GLM and NPAIRS.CVA software packages (or modules) using a novel second-level CVA. We found that for the block-design data, (a) slice timing correction and global intensity normalization have little consistent impact on the fMRI processing pipeline, but spatial smoothing, temporal detrending or high-pass filtering, and motion correction significantly improved pipeline performance across all subjects; (b) the combined optimization of spatial smoothing, temporal detrending and CVA model parameters on average improved between-subject reproducibility; and (c) the most important pipeline choices include univariate or multivariate statistical models and spatial smoothing. This study suggests that considering options other than simply using GLM with a fixed spatial filter may be of critical importance in determining activation patterns in BOLD fMRI studies.

Evaluation and comparison of GLM- and CVA-based fMRI processing pipelines with Java-based fMRI processing pipeline evaluation system.

Activation patterns identified by fMRI processing pipelines or fMRI software packages are usually determined by the preprocessing options, parameters, and statistical models used. Previous studies that evaluated options of GLM (general linear model)--based fMRI processing pipelines are mainly based on simulated data with receiver operating characteristics (ROC) analysis, but evaluation of such fMRI processing pipelines on real fMRI data is rare. To understand the effect of processing options on performance of GLM-based fMRI processing pipelines with real fMRI data, we investigated the impact of commonly-used fMRI preprocessing steps; optimized the associated GLM-based single-subject processing pipelines; and quantitatively compared univariate GLM (in FSL.FEAT and NPAIRS.GLM) and multivariate CVA (canonical variates analysis) (in NPAIRS.CVA)-based analytic models in single-subject analysis with a recently developed fMRI processing pipeline evaluation system based on prediction accuracy (classification accuracy) and reproducibility performance metrics. For block-design data, we found that with GLM analysis (1) slice timing correction and global intensity normalization have little consistent impact on fMRI processing pipelines, spatial smoothing and high-pass filtering or temporal detrending significantly increases pipeline performance and thus are essential for robust fMRI statistical analysis; (2) combined optimization of spatial smoothing and temporal detrending improves pipeline performance; and (3) in general, the prediction performance of multivariate CVA is higher than that of the univariate GLM, while univariate GLM is more reproducible than multivariate CVA. Because of the different bias-variance trade-offs of univariate and multivariate models, it may be necessary to consider a consensus approach to obtain more accurate activation patterns in fMRI data.

A Java-based fMRI processing pipeline evaluation system for assessment of univariate general linear model and multivariate canonical variate analysis-based pipelines.

As functional magnetic resonance imaging (fMRI) becomes widely used, the demands for evaluation of fMRI processing pipelines and validation of fMRI analysis results is increasing rapidly. The current NPAIRS package, an IDL-based fMRI processing pipeline evaluation framework, lacks system interoperability and the ability to evaluate general linear model (GLM)-based pipelines using prediction metrics. Thus, it can not fully evaluate fMRI analytical software modules such as FSL.FEAT and NPAIRS.GLM. In order to overcome these limitations, a Java-based fMRI processing pipeline evaluation system was developed. It integrated YALE (a machine learning environment) into Fiswidgets (a fMRI software environment) to obtain system interoperability and applied an algorithm to measure GLM prediction accuracy. The results demonstrated that the system can evaluate fMRI processing pipelines with univariate GLM and multivariate canonical variates analysis (CVA)-based models on real fMRI data based on prediction accuracy (classification accuracy) and statistical parametric image (SPI) reproducibility. In addition, a preliminary study was performed where four fMRI processing pipelines with GLM and CVA modules such as FSL.FEAT and NPAIRS.CVA were evaluated with the system. The results indicated that (1) the system can compare different fMRI processing pipelines with heterogeneous models (NPAIRS.GLM, NPAIRS.CVA and FSL.FEAT) and rank their performance by automatic performance scoring, and (2) the rank of pipeline performance is highly dependent on the preprocessing operations. These results suggest that the system will be of value for the comparison, validation, standardization and optimization of functional neuroimaging software packages and fMRI processing pipelines.